One-shot Face Recognition by Promoting Underrepresented Classes

@article{Guo2017OneshotFR,
  title={One-shot Face Recognition by Promoting Underrepresented Classes},
  author={Yandong Guo and Lei Zhang},
  journal={ArXiv},
  year={2017},
  volume={abs/1707.05574}
}
In this paper, we study the problem of training large-scale face identification model with imbalanced training data. This problem naturally exists in many real scenarios including large-scale celebrity recognition, movie actor annotation, etc. Our solution contains two components. First, we build a face feature extraction model, and improve its performance, especially for the persons with very limited training samples, by introducing a regularizer to the cross entropy loss for the multi-nomial… CONTINUE READING

Figures, Tables, Results, and Topics from this paper.

Key Quantitative Results

  • Our solution recognizes 94.89% of the test images at the precision of 99% for the one-shot classes.
  • With our feature extraction model and the UP term, we can recognize 94.89% of the test images in the low-shot set with a high precision of 99% and keep the top-1 accuracy of 99.8% for the base classes, while without using our method, only 25.65% of the test images from the low-shot set can be recognized at the same precision.
  • The evaluation results on the benchmark dataset show that the two new loss terms together bring a significant gain by improving the recognition coverage rate from 25.65% to 94.89% at the precision of 99% for one-shot classes, while still keep an overall accuracy of 99.8% for normal classes.

Citations

Publications citing this paper.
SHOWING 1-10 OF 32 CITATIONS

Incremental Classifier Learning with Generative Adversarial Networks

  • ArXiv
  • 2018
VIEW 9 EXCERPTS
CITES BACKGROUND & METHODS
HIGHLY INFLUENCED

Deep multi-factor forensic face recognition

VIEW 10 EXCERPTS
CITES BACKGROUND & METHODS
HIGHLY INFLUENCED

Feature Learning for One-Shot Face Recognition

  • 2018 25th IEEE International Conference on Image Processing (ICIP)
  • 2018
VIEW 7 EXCERPTS
CITES BACKGROUND & METHODS
HIGHLY INFLUENCED

One-Shot Face Recognition via Generative Learning

  • 2018 13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018)
  • 2018
VIEW 9 EXCERPTS
CITES BACKGROUND & METHODS
HIGHLY INFLUENCED

Doppelganger Mining for Face Representation Learning

  • 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)
  • 2017
VIEW 5 EXCERPTS
CITES METHODS & BACKGROUND
HIGHLY INFLUENCED

Face Generation for Low-Shot Learning Using Generative Adversarial Networks

  • 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)
  • 2017
VIEW 4 EXCERPTS
CITES METHODS & BACKGROUND
HIGHLY INFLUENCED

Learning Low-shot facial representations via 2D warping

  • ArXiv
  • 2017
VIEW 8 EXCERPTS
CITES METHODS, BACKGROUND & RESULTS
HIGHLY INFLUENCED

Low-Shot Face Recognition with Hybrid Classifiers

  • 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)
  • 2017
VIEW 5 EXCERPTS
CITES BACKGROUND & METHODS
HIGHLY INFLUENCED

References

Publications referenced by this paper.
SHOWING 1-10 OF 33 REFERENCES

Low-Shot Visual Recognition by Shrinking and Hallucinating Features

  • 2017 IEEE International Conference on Computer Vision (ICCV)
  • 2016
VIEW 6 EXCERPTS
HIGHLY INFLUENTIAL

Deep Residual Learning for Image Recognition

  • 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2015
VIEW 4 EXCERPTS
HIGHLY INFLUENTIAL

High Performance Large Scale Face Recognition with Multi-cognition Softmax and Feature Retrieval

  • 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)
  • 2017
VIEW 4 EXCERPTS
HIGHLY INFLUENTIAL

Low-Shot Face Recognition with Hybrid Classifiers

  • 2017 IEEE International Conference on Computer Vision Workshops (ICCVW)
  • 2017
VIEW 5 EXCERPTS
HIGHLY INFLUENTIAL

SphereFace: Deep Hypersphere Embedding for Face Recognition

  • 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2017
VIEW 4 EXCERPTS
HIGHLY INFLUENTIAL

Similar Papers